The Evolution of Genital Shape Variation in Female Cetaceans

ORIGINAL ARTICLE

doi:10.1111/evo.13395

The evolution of genital shape variation in female cetaceans

Dara N. Orbach,1,2,3 Brandon Hedrick,4 Bernd Wursig,¨ 5 Sarah L. Mesnick,6 and Patricia L. R. Brennan2,4 1Department of Biology, Dalhousie University, Life Science Center, 1355 Oxford Street, Halifax, NS B3H 4R2, Canada 2Department of Biological Sciences, Mount Holyoke College, Amherst, Massachusetts 3E-mail: [email protected] 4Department of Biological Sciences, University of Massachusetts-Amherst, Amherst, Massachusetts 5Department of Marine Biology, Texas A&M University at Galveston, Galveston, Texas 6Southwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, La Jolla, California

Received June 28, 2017 Accepted November 5, 2017

Male genital diversification is likely the result of sexual selection. Female genital diversification may also result from sexual selection, although it is less well studied and understood. Female genitalia are complex among whales, dolphins, and porpoises, especially compared to other vertebrates. The evolutionary factors affecting the diversity of vaginal complexity could include ontogeny, allometry, phylogeny, sexual selection, and natural selection. We quantified shape variation in female genitalia using 2D geometric morphometric analysis, and validated the application of this method to study soft tissues. We explored patterns of variation in the shape of the cervix and vagina of 24 cetacean species (n = 61 specimens), and found that genital shape varies primarily in the relative vaginal length and overall aspect ratio of the reproductive tract. Extensive genital shape variation was partly explained by ontogenetic changes and evolutionary allometry among sexually mature cetaceans, whereas phylogenetic signal, relative testis size, and neonate size were not significantly associated with genital shape. Female genital shape is diverse and evolves rapidly even among closely related species, consistent with predictions of sexual selection models and with findings in invertebrate and vertebrate taxa. Future research exploring genital shape variation in 3D will offer new insights into evolutionary mechanisms because internal vaginal structures are variable and can form complex spirals.

KEY WORDS: Allometry, cetacean, evolution, geometric morphometrics, natural selection, ontogeny, phylogeny, sexual selection, vagina.

Research on genital morphology has largely focused on exploring adaptations can be subtle, careful quantification of morpholog- the often extreme morphological variation found in male intro- ical attributes is critical to identify patterns and the underlying mittent organs, with female genitalia considered less variable than evolutionary drivers of variation (Brennan and Prum 2015). males and subsequently not as well studied (Ah-King et al. 2014; Morphological variation in female genitalia, sometimes as- Brennan 2016). This oversight is surprising as morphological sessed concurrently with male genital variation, may reflect sex- variation in male and female genitalia should covary because of ual selection pressures operating during mating via female mate close mechanical interactions during copulation (Brennan and choice, intrasexual male competition, or sexual conflict (Hosken Prum 2015). Recent studies have quantified previously unreported and Stockley 2004; Simmons 2014; Brennan and Prum 2015). In variation in female genital morphology in several taxa, validating addition, morphological variation in female genitalia may also be the value of exploring female reproductive tract morphology influenced by natural selection that prevents interspecific mating (flies, Puniamoorthy et al. 2010; Yassin and Orgogozo 2013; (lock and key hypothesis; Masly 2012), ensures successful ovipo- waterfowl, Brennan et al. 2007; watersnakes, Showalter et al. sition, or facilitates an easy parturition of neonates or egg-laying 2013; cetaceans, Orbach et al. 2017a). Because female genital (Brennan 2016). Female vaginal shapes and structures can also

C 2017 The Author(s). Evolution C 2017 The Society for the Study of Evolution. 1 Evolution DARA N. ORBACH ET AL.

exhibit ontogenetic variation, as reproductive organs are often predicted to influence cetacean genital shape as interpopulation not functional at birth in vertebrates and change into their adult variation in testes mass and mating systems are known to occur form during sexual maturation (e.g., rats, Berdnikovs et al. 2007; in spinner dolphins (Stenella longirostris; Perrin and Mesnick snakes, Showalter et al. 2013). 2003). Despite findings that vaginal length scales isometrically to Cetaceans (whales, dolphins, and porpoises; 90 extant body length in cetaceans, we predicted that evolutionary allome- species) are a speciose clade of mammals that evolved from a try would affect genital shape variation because of the positively common terrestrial ancestor. Whales, dolphins, and porpoises ex- allometric relationship between vaginal folds and body length hibit a diverse array of adaptations to aquatic living (reviewed in (Orbach et al. 2017a). We also predicted that shared evolutionary Berta et al. 2015), including social organizations ranging from history would affect genital shape because a moderate phyloge- solitary in some large baleen whales to highly gregarious in some netic signal was reported between vaginal length and body length small toothed whales. Their diverse life histories and adaptations, across the cetacean clade (Orbach et al. 2017a) and morpholog- coupled with the unusual genital complexity of females, make ical traits are expected to be more similar among closely related cetaceans a particularly interesting group to investigate patterns species (e.g., Kamilar and Cooper 2013). Sexual and natural selec- of morphological evolution. Recent work on cetaceans has doc- tion can influence genital shape (e.g., Hosken and Stockley 2004; umented widespread variation in female genital structures called Langerhans et al. 2005; Reinhardt 2010; House et al. 2013). Using vaginal folds, which appears to be unparalleled in other mam- testes size as a proxy for postcopulatory competition, we predicted malian taxa (Orbach et al. 2017a). With the exception of ar- that species with larger relative testes (higher sperm competition) tiodactyls, the closest relatives to cetaceans, other mammalian may have longer and narrower vaginas to make insemination groups do not have vaginal folds (Pabst et al., 1998). These folds more difficult. Cetaceans have larger testes-to-body size ratios are muscular protrusions of the vaginal wall into the lumen, and than similar-sized terrestrial mammals, suggesting they engage their function remains unclear (reviewed in Clarke et al. 1994; in sperm competition (Kenagy and Trombulak 1986; Aguilar and Orbach et al. 2017a). The vaginal folds of cetaceans vary in num- Monzon 1992), although there is extensive interspecific variation. ber, thickness, positioning, shape, and size across species and ap- In addition, we predicted that neonate size may influence vaginal pear to be under strong selection pressures because fold lengths shape, so that relatively larger neonates may be associated with a are positively allometric with body size (Orbach et al. 2017a). shorter and wider vagina to facilitate parturition. Finally, because However, quantifying vaginal fold attributes is difficult, as mor- vaginal shape may be under selection from both of these traits phological characteristics may vary within a species (Orbach et al. simultaneously (Brennan and Prum 2015), we examined the in- 2016, 2017a). In addition to these folds, cetacean genitalia vary teraction between relative testes mass, relative neonate size, and across species in the relative length, width, and proportions of the genital shape. cervix and vaginal regions (Fig. 1), herein referred to broadly as genital shape. However, this variation has yet to be systematically described and examined. Material and Methods Studies of shape variation have revolutionized our under- DATA COLLECTION standing of morphological evolution, particularly concerning Reproductive tracts were obtained opportunistically from fresh complex structures. Several studies have used either Fourier anal- (<24 h postmortem) or moderately decomposed deceased female ysis or geometric morphometric approaches to examine shape cetaceans. Sexually immature and mature specimens were col- variation in insect genitalia (Arnqvist 1998; Garnier et al. 2005; lected from marine mammal response networks and research insti- Pizzo et al. 2006; Simmons and Garcia-Gonzalez 2011), and more tutions in the United States, New Zealand, and South Africa under recently, vertebrate genitalia (Evans et al. 2011; Heinen-Kay and government authorization (see Ethics statement). Most specimens Langerhans 2013; Showalter et al. 2013; Simmons and Firman were collected from toothed whales because of their smaller sizes 2014). However, quantitative studies of female genital shape vari- compared to baleen whales. The marine mammal stranding net- ation have not been previously reported in mammals. works provided information on the total body length (Norris 1961) We use a geometric morphometric approach to analyze 2D and sexual maturity state of specimens (based on regional asymp- genital shape variation in female cetaceans. We examine several totic body lengths or presence of corpora albicantia or corpora hypotheses that may explain genital shape variation including on- lutea on the ovaries; Perrin and Donavan 1984). togeny, geography, evolutionary (ontogenetic-controlled) allom- Intact reproductive tracts (from the ovaries to the external etry, phylogeny, testes mass, and neonate size. We predicted on- urogenital slit) were excised from the postmortem animals, frozen togeny would influence the shape of cetacean genitalia and reveal immediately, and transferred to necropsy facilities. Specimens which dimensional aspects are most important for copulation or were defrosted and oriented in a dorsal recumbency. A single parturition as animals become sexually mature. Geography was longitudinal incision was made consistently down the ventral

2 EVOLUTION 2017 EVOLUTION OF FEMALE CETACEAN GENITAL SHAPE

Figure 1. Cetacean genital shapes mapped onto a phylogenetic tree. Phylogeny showing the cetacean taxa examined in our analysis with the corresponding female reproductive tract shape. The cervix is oriented on the right side and the vaginal opening is on the left side. All genital shapes were subjected to general Procrustes analysis. In species with more than one specimen, a consensus (mean) shape is presented. Species in which there were only sexually immature specimens are red, only sexually mature specimens are blue, and a combination of sexually immature and mature specimens are purple. A labeled reproductive tract is depicted including the order that the landmarks were applied across specimens. midline, from the clitoris through to the internal bifurcation of the vagina (Orbach et al. 2016). To quantitatively capture variation in uterine horns (Orbach et al. 2016). To further ensure the incision genital shape across a range of species and body sizes, each repro- was exactly midline through the ventral wall of the reproductive ductive tract was photographed in a standardized bird’s-eye view tract, the incision was consistently made through the midpoint of using digital cameras with a minimum resolution of 10.1 megapix- the bladder (connected to the ventral vaginal wall). The vaginal els. The focal plane of the camera was positioned parallel to the walls were splayed open to reveal the contours of the cervix, vagi- midline of the reproductive tract, with the long axis of the vaginal nal folds, and clitoris. As there can be extensive asymmetry within canal oriented such that the vaginal opening was at the bottom of the cetacean reproductive tract, specific landmarks found on the the image. Scales were positioned in both transverse and coronal dorsal midline of reproductive tracts were aligned in a straight (frontal) planes. Given that the uterine horns’ positioning was not line to ensure consistent orientation for measurements and digi- a primary concern of this study, they were not digitized and their tization. Specifically, the internal bifurcation of the uterine horn positioning was not oriented during photography. Although soft was aligned in the same straight line as the cranial limit of the tissue is harder to align in a repeatable manner compared to hard

EVOLUTION 2017 3 DARA N. ORBACH ET AL.

tissue, the vaginal structure was robust and maintained its shape Singleton (2002) and for each individual specimen using the during imaging (see error analysis next). Because the tissues were method described by Lockwood et al. (2002). We landmarked collected from recently postmortem specimens and stored frozen, a single sexually mature common bottlenose dolphin (Tursiops they were not subjected to the shrinkage typically in association truncatus) specimen not used in any other analysis (specimen ID: with formalin or ethanol preservation (Fox et al. 1985). MARS2017-137) 11 times to create an error dataset (Tables S1 and S3). This specimen was remounted prior to taking each pho- GEOMETRIC MORPHOMETRIC ANALYSES tograph to simulate the error inherent in the mounting process and 2D geometric morphometric techniques were used to capture the in the photography process as a whole, in addition to error in the genital shape of the cervix, cranial vagina, and caudal vagina. We landmarking process. selected a geometric morphometric approach over measuring the We performed a GPA on the error dataset, calculated the more traditional Euclidean distances because we were interested consensus shape using the gpagen and mshape functions in the in describing how the overall outline shapes varied rather than R package geomorph (Adams and Otarola-Castillo,´ 2013), and how individual landmarks varied with respect to one another. exported the landmark data into Microsoft Excel version 12.3.3 Photographs of reproductive tracts were imported into tpsDIG2 (Microsoft, 2008). The details of the GPA analysis can be found (Rohlf 2006; Fig. S1), and landmarks were used to outline the in Table S1. After calculating error for individual landmarks, we cranial and caudal bounds of the cervix, largest vaginal fold, and included the error sample subset in the original dataset to qual- vaginal opening (Figs. 1 and S2). We traced the shape of the itatively evaluate how the error subset plotted in morphospace. borders of these three regions using 10 semilandmark curves with This allowed us to evaluate whether there was substantial scatter 128 semilandmarks (Fig. S2). The cervix was delineated from the in our error dataset relative to other specimens. Given that many cranial vaginal region by the caudal end of the ectocervix (portio of our assertions were based on Principal component analysis vaginalis; Fig. S2). The cranial vaginal region was delimited from (PCA), this was a valuable check on our data capture process. Fi- the caudal vaginal region by the largest vaginal fold (the fold of nally, following Lockwood et al. (2002) and Hedrick and Dodson greatest protrusion into the vaginal lumen from the vaginal wall; (2013), we calculated the Euclidean distances between PC1, 2, 3, Fig. S2). The caudal demarcation of the caudal vaginal region and 4 (PC variance > 5%) for both nonerror specimens and error was the cranial limit of the vulva; a natural change in tissue color specimens and the mean of the error sample for each PC. This was used to extend the transverse line from the cranial limit of allowed us to quantitatively evaluate the amount of scatter of the the vulva laterally to the bisected clitoris (Orbach et al. 2016; error subset within the total specimen morphospace. Fig. S2). We included the cervix in our analysis as it may play a role in copulation because the penis tip penetrates the cervix in OVERALL GENITAL SHAPE some artiodactyls (Bravo et al. 1996). Overall geometric morphometric patterns of the cervix and vagina Landmark configurations were imported into R (R Core De- were visualized by overlaying the reproductive tract shapes sub- velopment Team 2017) and subjected to general Procrustes anal- jected to GPA onto a cetacean phylogenetic tree (McGowen et al. ysis (GPA) using the R package geomorph (Adams and Otarola-´ 2009). A consensus (average) genital shape was used when there Castillo 2013). GPA translates, rescales, and rotates all landmark was more than one specimen per species. PCAs were run on all configurations into a common orientation (Zelditch et al. 2012). cetacean specimens to visualize general genital shape trends in Semilandmark curves were slid such that bending energy between the data. semilandmarks was minimized (Perez et al. 2006). Although we found substantial asymmetry in the genitalia, the asymmetry is bi- ONTOGENETIC ALLOMETRY ological and unrelated to mounting, landmarking error, or preser- Intraspecific ontogenetic and geographic patterns were explored vation (see error analysis next). Therefore, we chose to use the using common bottlenose dolphins (T. truncatus) and harbor por- raw asymmetric data rather than correcting for asymmetry us- poises (Phocoena phocoena). We had large representative samples ing the symmetrical component of variation because the latter of sexually immature and mature specimens and different regions would eliminate potentially important information relating to true of stranding for these two species. The ontogenetic and geographic biological shape. allometric analyses were run separately for common bottlenose dolphins and harbor porpoises using log-transformed centroid ERROR ANALYSIS size and the common allometric component (CAC) method de- To reduce landmark error and eliminate interobserver error, all veloped by Mitteroecker et al. (2004). Centroid size is the squared specimen photography and landmarking were performed by one root of the sum of the square distances of a set of landmarks from researcher (DNO). Intraobserver landmark error was calculated their centroid (Bookstein 1991). The CAC is a vector of regression for each individual landmark using the method described by scores generated by a pooled regression of shape variables on size,

4 EVOLUTION 2017 EVOLUTION OF FEMALE CETACEAN GENITAL SHAPE

corrected for species means. Our application of this method to two cetacean species is similar to that used by Drake and Klingenberg (2008) for single taxa. Based on the results of the ontogenetic allometric analyses, all subsequent tests were restricted to inclusion of sexually mature cetaceans only.

EVOLUTIONARY ALLOMETRY We examined evolutionary allometric relationships to determine if body size and vaginal size in sexually mature cetaceans influence genital shape. Evolutionary allometry was assessed by plotting the CAC against the log-transformed centroid size of the landmark configurations. This approach was repeated by plotting the CAC against the log-transformed total body length of the specimens.

PHYLOMORPHOSPACE The genital shape morphospace of sexually mature specimens was explored using a PCAs. We then examined the effect of phylogeny on our data using a recent time-calibrated cetacean tree (McGowen et al. 2009) by calculating the K-test statistic developed by Blomberg et al. (2003) and modified for high-dimensional morphometric data by Adams (2014). The mean genital shape was calculated for species with more than one sexually mature specimen following Sherratt et al. (2014). The K-statistic was Figure 2. Error analysis demonstrating limited intraspecimen analyzed using the physignal function in geomorph (Adams and measurement error. (A) Principal component analysis of complete Otarola-Castillo´ 2013). A phylogeny was overlaid on the PC mor- cetacean sample (n = 61) and error subset sample (n = 11). This phospace to visualize the relationship between phylogeny and demonstrates that in PC morphospace, members of the error sub- genital shape using the plotGMPhyloMorphoSpace function in set cluster more closely to other members of the error subset geomorph (Adams and Otarola-Castillo´ 2013). than to other specimens. (B) Distribution of Euclidean distances between each specimen and the mean of the error subset dataset. TESTES MASS AND NEONATE SIZE Error samples were more similar to one another than any was to We assessed the effects of maximum testes mass and average nonerror samples. neonate size on genital shape variation. The residuals of maximum combined left and right testes mass and maximum sexually typically found in hard tissue analyses (e.g., Foth et al. 2016). We mature male mass (g), and the residuals of average neonate body then examined whether our error sample substantially diverged length at birth and average mother body length at birth (cm) qualitatively and quantitatively in morphospace using Euclidean were calculated based on data obtained from an extensive litera- distances between specimens and the mean of the error sample. ture review (Table S2). Two species (Mesoplodon peruvianus and We found that all error specimens overlapped in the same region Mesoplodon stejnegeri) were excluded from the analysis because of morphospace and that no other specimens overlapped with the we were unable to find published data on testes mass for these error sample (Fig. 2). The Euclidean distances demonstrated that species. We ran a phylogenetic general least squares (pgls) model all error specimens were more similar to one another than to any in the R package caper (Orme 2013) and evaluated the effect of nonerror specimens (Table S3; Fig. 2). both testes and neonate residuals and their interaction on vaginal shape. OVERALL GENITAL SHAPE Sixty-one reproductive tracts were obtained, representing 24 species and seven families of marine mammals (Table S4; Fig. S1). Results Accordingly, our data have representation from 26% of extant ERROR ANALYSIS cetacean species (McGowen et al. 2009), with the diversity of We found that error ranged between 2.22% (landmark 8) and reproductive tract shapes depicted in Figures 1 and S1. However, 7.77% (landmark 4), with a mean percent error of 5.11% only one baleen whale specimen was obtained. In the PCA of all (Table S1). This error was relatively small, but larger than that specimens, principal component 1 (PC1) and PC2 accounted for

EVOLUTION 2017 5 DARA N. ORBACH ET AL.

Figure 3. Principal components analysis of cetacean genital shapes for all specimens (n = 61). Common bottlenose dolphin (Tursiops truncatus) and harbor porpoise (Phocoena phocoena) specimens are highlighted to demonstrate the extent of intraspeciﬁc variation driven by ontogenetic allometry. Sexually immature specimens are represented by squares, whereas sexually mature specimens are represented by circles. Specimen numbers in the plot refer to specimens in Table S4.

69.1% of the total variance (Table S4). As the remaining PCs after Sexually mature females have wide cervices, but vary in the rela- PC2, each accounted for less than 10% of the variance (Table S4), tive size of the cervix and cranial vagina. Sexually immature fe- only the patterns found in PC1 and PC2 are described. PC1 ex- males vary in the relative width of the cervix and relative length of plained 54.6% of the total genital shape variation and was driven the caudal vagina. The extensive degree of intraspecific variation by the relative length of the cranial and caudal vaginal regions in reproductive tract shape, largely attributed to ontogenetic fac- (Fig. 3). PC2 explained 14.5% of the total genital shape variation tors, is highlighted in Figure 3, which also emphasizes that speci- and was driven by a change in the overall aspect ratio (the ratio mens for both T. truncatus and P. phocoena are found throughout of width to height) of the cervix and vagina (Fig. 3). the morphospace for cetaceans as a whole. Based on the clear effects of ontogenetic allometry on cetacean genital shape, only sexually mature specimens were used in subsequent analyses. ONTOGENETIC ALLOMETRY The effects of population could not be disentangled from the ef- Ontogeny had a significant effect on variation in female reproduc- fects of body size among our samples because sexually immature tive tract shape in common bottlenose dolphins (T. truncatus; n = specimens were coincidentally obtained primarily from different = = 14, F12 2.52, P 0.027) and in harbor porpoises (P. phocoena; populations than sexually mature specimens (Fig. 4E and F). n = 14, F12 = 8.05, P < 0.001), the two species for which we had a large enough sample size to explore ontogenetic trends. The relative width of the cervix and vagina change with size (Fig. 4A EVOLUTIONARY ALLOMETRY and B), as does the relative length of the cranial vagina among Thirty-one of the 61 specimens were sexually mature and included harbor porpoises (Fig. 4B). However, ontogeny is a stronger pre- in the analysis of evolutionary allometry and phylogeny. Evolu- dictor of genital shape in P. phocoena (R2 = 0.402) compared tionary allometry was a significant predictor of reproductive tract to T. truncatus (R2 = 0.174). PC1 divided reproductive tract shape using the log-transformed total body length of specimens 2 shapes distinctly by reproductive state in P. phocoena (Fig. 4D; (F29 = 8.07, P < 0.001, R = 0.21), but not the log-transformed 2 Table S5), where sexually mature specimens had wider genitals centroid size of the vagina (F29 = 2.15, P = 0.095, R = 0.07; overall, and the cranial vaginal region was the most prominent. In Table S7). Reproductive tracts were overall narrower and with T. truncatus, sexually mature females vary along PC1, whereas longer caudal vaginal regions as specimens increased in body sexually immature females vary along PC2 (Fig. 4C; Table S6). length (Fig. 5).

6 EVOLUTION 2017 EVOLUTION OF FEMALE CETACEAN GENITAL SHAPE

Figure 4. Within species ontogenetic allometry of cetacean genital shapes. Ontogenetic trajectory plotting genital shape against centroid size with regression line for (A) common bottlenose dolphins (Tursiops truncatus) and (B) harbor porpoises (Phocoena phocoena). Shape trends divided into sexually immature (black) and sexually mature (red) specimens within (C) T. truncatus and (D) P. phocoena. (E) Shape trends splitting T. truncatus specimens into populations (North Carolina—red; Texas—green; Florida—black; Virginia—blue). (F) Shape trends splitting P. phocoena specimens into populations (California—red, Massachusetts—green, Alaska—black, Oregon—blue).

PHYLOMORPHOSPACE TESTES MASS AND NEONATE SIZE PC1 explained 72.7% of the morphospace variance in the repro- The residuals of maximum combined left and right testes mass ductive tract shape of sexually mature species (n = 17) and shows to maximum male body mass did not significantly affect female 2 a narrowing and lengthening of the caudal vaginal region (Fig. 6; genital shape (F13 = 1.47, P = 0.62, R = 0.10). Similarly, the Table S8). Phylogeny was not a significant predictor of reproduc- residuals of average neonate to mother length at birth did not have tive tract shape among sexually mature cetaceans (Kmult = 0.125, a significant effect on reproductive tract shape variation (F15 = P = 0.821). 2.54, P = 0.642, R2 = 0.14). The interaction between relative

EVOLUTION 2017 7 DARA N. ORBACH ET AL.

Figure 5. Evolutionary allometry of cetacean genital shapes. (A) Evolutionary allometric trajectory between shape (CAC) and the log- transformed centroid size of the vagina and (B) between shape (CAC) and log-transformed body length with regression line.

Figure 6. Phylomorphospace of cetacean genital shapes. Visualization of the morphospace of sexually mature specimens with phylogeny overlain. Colors indicate monophyletic clades on inset cladogram. testes mass and relative neonate size also did not significantly comparative measurements have been limited to linear dimen- affect female reproductive tract shape variation (F3 = 3.34, P = sions (Orbach et al. 2017a). As shape can be a large component 0.243, R2 = 0.36; Table S2). of morphological evolution and provide insights into functionality (Evans and Sanson 2003), the application of a 2D geometric morphometric approach enabled us to quantitatively analyze fine- Discussion scale morphological shape variation and explore potential evolu- Our research examined the evolution of female genital shape in a tionary drivers. Our results are consistent with previous studies clade where extensive genital diversity has been documented but in several taxa, where extensive variation in female genitalia has

8 EVOLUTION 2017 EVOLUTION OF FEMALE CETACEAN GENITAL SHAPE

been reported, underscoring the need for more careful quantifica- that may be associated with the role of the penis tip in sperm tion of female morphological variation (reviewed in Brennan and deposition proximate to the cervix. In both P. phocoena and T. Prum 2015). For example, in insects, the shape of female geni- truncatus, the penis shaft stops at the largest vaginal fold that talia has been shown to evolve rapidly (e.g., flies; Puniamoorthy divides the vagina into the cranial and caudal regions, while the et al. 2010), and females can be more variable in shape than penis’ filiform tip can continue cranially in the vagina (Orbach males (Scarab beetles, Pyllophaga hirticula; Polihronakis 2006). et al 2017b). Therefore the shape change in the cranial vagina Even when male genital shape changes faster than female shape, may accommodate movement of the penis tip. As ontogeny was a female shape can be extremely variable (stink bugs; Genevcius stronger predictor of genital shape in P. phocoena compared to T. et al. 2017). In cases where morphological variation in female truncatus, we suggest ontogenetic effects may vary with the ex- genitalia are apparently subtle, careful quantification of shape tent of vaginal complexity. In T. truncatus, there is typically only variation has shown significantly different vaginal shapes despite one fold, but in P. phocoena there are several folds concentrated some overlap in morphospace (watersnakes, Nerodia sipedon and mainly in the cranial region of the vagina (Orbach et al. 2017a). N. fasciata; Showalter et al. 2013), and significant morphological The allometric relationship suggests that reproductive tract shape differences (Yassin and Orgogozo 2013). Evidence that female changes occur gradually, rather than being punctuated by events genital shape evolves rapidly and covaries with male genital vari- leading to vaginal and cervical distention, such as first sexual ation is well established in several taxa (reviewed in Brennan and intercourse or first parturition (Fig. 4A and B). Prum 2015), and our study provides a first step in investigating this The cervices of T. truncatus and P. phocoena also widen pattern more broadly in mammals, where no comparative studies as females become sexually mature, consistent with reports of of female genital shape across a clade have been conducted. ontogenetic changes in tissue composition related to the func- In general, the genital shape of cetacean reproductive tracts tional role of the cervix in reproduction in various mammalian varies primarily on the relative length of the caudal and cranial taxa (El-Banna and Haffez 1972; Kress and Mardi 1992). To our vagina, followed by some variation in the aspect ratio of the vagina knowledge, this is the first study to show an ontogenetic change and cervix. Ontogenetic factors largely contribute to the overall in the shape of the cervix. In alpaca (Vicugna pacos), the tip of intraspecific genital shape variation (Fig. 3). The separation be- their fibroelastic penis penetrates through the cervix (Bravo et al. tween the cranial and caudal vagina was delimited by the largest 1996). However, the penis tip in cetaceans is unlikely to reach all vaginal fold. This geometrically homologous separation was a the way to the cervix even at full distention (Orbach et al. 2017b). consistent and obvious landmark across specimens. Further, the largest fold is under strong mechanical and likely sexual selection GEOGRAPHY pressure; the largest fold stops progression of the penis shaft dur- We predicted geography would influence cetacean genital shape. ing simulated intromission, while the penis tip likely continues However, the roles of geographic and ontogenetic allometry on into the cranial vagina (Orbach et al. 2017b). Therefore over- reproductive tract shape were confounded in our data, as sex- all vaginal shape variation is likely to be associated with penis ually immature specimens were coincidentally collected from morphology in cetaceans. different populations than sexually mature specimens and ontogeny accounted for a large percentage of overall shape variation ONTOGENY (Fig. 4C and D). The opportunistic nature of specimen acquisition Ontogeny significantly influences the genital morphology of com- hindered our ability to collect specimens of both maturity states mon bottlenose dolphins (T. truncatus) and harbor porpoises (P. from the same populations. Although geographic divergence and phocoena; Fig. 4). As our results differ from findings that used associated natural selection pressures can influence male genital linear measurements to explore ontogenetic changes in the geni- morphology (e.g., landsnail, Helix aspersa; Madec and Guiller talia of T. truncatus (Orbach et al. 2016), we advocate the value 1994; guppies, Poecilia reticulata; Evans et al. 2011; deer mice, of using multiple analytical tools to explore patterns of variation Peromyscus maniculatus and Peromyscus oreas; Sullivan et al. in morphology. Ontogenetic genital shape changes have the po- 1990), geographic effects on female genitalia have not been well- tential to reveal which aspects of genitalia are most important to documented and further research is warranted. perform a copulatory or reproductive function. For example, sexually mature watersnakes have a bifurcation in their vaginas that EVOLUTIONARY ALLOMETRY AND PHYLOGENETIC is not well-developed in sexually immature females, and males PATTERNS have bifurcated hemipenes, suggesting that the bifurcation of fe- Evolutionary allometry was a significant predictor of female gen- males is necessary for copulatory function (Showalter et al. 2013). ital shape in cetaceans (Fig. 5), just as it was a predictor of vaginal Similarly, in this study, we observed an extension and broadening length (Orbach et al 2017a). However, vaginal size was not asso- of the cranial vagina in sexually mature females (Fig. 4C and D) ciated with vaginal shape. Accumulating evidence suggests that

EVOLUTION 2017 9 DARA N. ORBACH ET AL.

genital shape, not just vaginal length, should be considered in studies that assess the relationship between testes mass and penis analyses of genital scaling patterns. Size and shape components shape may yield interesting results. As vaginal shape morphology are important considerations for the functional morphology of did not change relative to neonate size, shape may not be influ- genitalia, and both can vary allometrically with body size. For enced by natural selection pressures associated with parturition. example, the shape of both male and female genitalia varies more rapidly than genital size (e.g., Ontophagus beetles, Macagno et al. ERROR ANALYSIS 2011; O. taurus, Simmons et al 2009; Mus musculus, Simmons Intraobserver landmarking error was small (5.11%). However, and Firman 2014), suggesting that changes in size are more con- because error is calculated based on a ratio involving the dis- strained than changes in shape; this may facilitate mechanical fit tance from each landmark to the consensus centroid, landmarks during copulation (reviewed in Brennan and Prum 2015). closer to the centroid have higher error (von Cramon-Taubadel The lack of a phylogenetic signal in our data (Figs. 1 and 6; et al. 2007). This could explain why landmarks 3, 4, and 5, which Table S8) suggests that vaginal shape evolves rapidly even among were closest to the centroid, had the highest error. Regardless, closely related species, some of which had very different genital even the error surrounding these landmarks is small relative to shapes (Figs. 1 and 5). A high degree of morphological divergence interspecimen differences (Fig. 2). Singleton’s (2002) method is among closely related species is predicted by the “lock and key” appropriate as it quantifies error for each landmark individually mechanism of genital evolution (Masly 2012), but is not incon- and thus we considered its advantages to outweigh its disadvan- sistent with other models (Brennan and Prum 2015). However, tages in the present analysis. Soft tissue analyses using geometric in the absence of studies of penile morphology, it is impossible morphometric approaches are rare and may be more prone to to distinguish the evolutionary mechanism responsible for this error than hard tissue analyses, so we examined the position of rapid divergence (Brennan and Prum 2015). In at least P. pho- the error dataset in morphospace and quantified the Euclidean caena and T. truncatus, sexual conflict may partially drive genital distances within and between the error sample and nonerror spec- evolution because the largest vaginal fold acts as a barrier to fur- imens. Given that there was no overlap in total shape between ther intromission of the penis shaft into the vagina (Orbach et al the error sample and other specimens either qualitatively in mor- 2017b). Vaginal barriers to penile penetration have been shown to phospace (Fig. 2), no overlap in Euclidean distances (Table S3, result from sexual conflict in waterfowl (Brennan et al 2007). Fig. 2; Lockwood et al. 2002), and that individual landmark error Intraspecific variation in shape was evident in our samples, was small (Table S1; Singleton 2002), we assert that error did not although our samples sizes were too small to quantify this po- play a large role in our analysis and that the results demonstrating tentially confounding factor. Genital shape can be highly variable large within species shape divergence are real trends. This also within species (e.g., scarab beetles, Polihronakis 2006; water- suggests that the use of geometric morphometric techniques to snakes, Showalter et al. 2013). We eliminated all sexually imma- quantify soft tissue shape can provide repeatable data, validating ture specimens from the evolutionary allometry and phylogeny its potential application to other soft tissue systems. analyses to minimize the effects of ontogenetic variation, but Although our data include 26% of all extant cetacean species other aspects of reproductive state, such as pregnancy and partu- and could be expanded upon further, we believe our data are rition, may affect vaginal shape (e.g., Homo sapiens; Pendergrass robust and representative of taxon-wide patterns. A larger sam- et al. 2000). ple size per species could further elucidate changes that may be associated with reproductive state. For example, vaginas are TESTES MASS AND NEONATE SIZE longer in pregnant compared to lactating and resting T. trunca- We found no significant correlations between genital shape vari- tus; however, sample sizes were too small to draw biologically ation and relative testes weight, relative neonate size, or their relevant conclusions (Orbach et al. 2016). Our data included only interaction. Although residual testes size is a good predictor of one baleen whale species. Due to the larger sizes of baleen whales the intensity of sexual selection (Kenagy and Trombulak 1986), compared to toothed whales, obtaining baleen whale reproductive the vagina interacts directly with the penis during copulation and tracts is an obstacle that should be overcome in future research therefore penile morphology may be a better predictor of vaginal to explore the potential roles of additional life-history factors in shape than testes size (e.g., waterfowl, Brennan et al. 2007). Fu- genital morphology evolution. We also acknowledge that repro- ture studies that assess the coevolution of penis length and vaginal ductive tract shape variation could reflect genetic drift rather than shape in cetaceans at the interspecific level may find that sexual an adaptive function(s). However, because reproductive structures selection influences female genital shape evolution through penis are immediately and critically important to an organisms’ fit- shape directly. Cetacean species with large relative testes masses ness, nonadaptive deviations could have devastating consequences have comparatively long penises (Brownell and Ralls 1986; Dines for an individual. Alternatively, vaginal complexity may be un- et al. 2014), as do rodents and carnivores (Ramm 2007). Future der selection, and genital shape variation is a by-product that

10 EVOLUTION 2017 EVOLUTION OF FEMALE CETACEAN GENITAL SHAPE

accommodates more vaginal folds. If so, we would expect to find K. Danil, A. Evans, and two anonymous reviewers for their insightful most variation in the cranial vaginal region, where most folds are edits. This research was supported by grants to DNO from Texas A&M concentrated (Orbach et al. 2017a), but our results here do not University at Galveston (Department of Marine Biology; Texas Institute of Oceanography), American Museum of Natural History (Lerner Gray support that hypothesis. The folds form a complex 3D structure in Memorial Fund), Natural Science and Engineering Research Council of the female vaginal lumen (Orbach et al. 2017b), and therefore 3D Canada (PGSD2-420080-2012), and Texas Sea Grant (Grands-In-Aid of morphometric analyses that incorporate spiraling patterns may Graduate Research). This work was also supported by a faculty grant from better capture the complexity and extensive variation of female Mount Holyoke College and the University of Massachusetts Amherst to PLBR, and from Texas A&M University at Galveston (Department of genitalia among cetaceans. For example, 3D casts of the vagi- Marine Biology; George Mitchell Chair in Sustainable Fisheries) to BW. nal lumen in humans have shown variation in shape, length, and The authors declare they have no competing interests. width that may correlate with ethnicity and parturition history (Pendergrass et al. 2000). We recommend that reproductive tracts Ethics: Specimens in the United States of America were collected be collected and preserved intact to explore 3D genital shape under National Marine Fisheries Service (NMFS) salvage permit patterns so that larger sample sizes can be obtained. letters to DNO. Specimens from New Zealand were imported to In conclusion, ontogeny and evolutionary allometry explain the United States of America under an institutional Convention patterns of genital shape variation in cetaceans, while phylogeny, on International Trade in Endangered Species of Wild Fauna and testes size, and neonate size do not. Rapid evolution of female Flora permit (CITES; 14US690343/9). Specimens from South genital shape, as suggested by the lack of phylogenetic signature Africa were authorized by the South African Departments of En- reported here, seems to be a common feature of female genitalia vironmental Affairs and Agriculture, Forestry, and Fisheries to across taxa and suggests an important role for sexual selection in the Port Elizabeth Museum/Bayworld (RES2016-62). Specimens their evolution. Ontogenetic patterns of genital variation may be in South Africa were photographed on-site and not exported. This a widespread feature of vertebrate genitalia. Future efforts that study was exempt from an Institutional Animal Care and Use describe 3D shape variation and analyze male morphology are Committee (IACUC) authorization as the specimens were de- likely to be fruitful in further explaining the complexity of female ceased upon acquisition and the salvage materials were authorized genitalia among cetaceans. from appropriate government agencies.

AUTHOR CONTRIBUTIONS LITERATURE CITED PLRB conceived the project idea. BW and PLRB funded the specimen Adams, D. C. 2014. A generalized K statistic for estimating phylogenetic collection. DNO and SLM collected and dissected the specimens. DNO signal from shape and other high-dimensional multivariate data. Syst. traced the photographs in TPS. BH wrote the statistical code, ran the Biol. 63:685–697. analyses, and created the figures. DNO wrote the initial manuscript Adams, D. C., and E. Otarola-Castillo.´ 2013. Geomorph: an R package for the with feedback from coauthors. All authors contributed toward manuscript collection and analysis of geometric morphometric shape data. Methods revisions. Ecol. Evol. 4:393–399. Aguilar, A., and F. Monzon. 1992. Interspecific variation of testis size in ACKNOWLEDGMENTS cetaceans: a clue to reproductive behaviour. Eur. Res. Cetaceans 6:162– We thank Alaska Veterinary Pathology Services (particularly K. Burek 164. and the J. P. Prescott grant: NA15NMF4390053. NOAA MMHSRP per- Ah-King, M., A. B. Barron, and M. E. Herberstein. 2014. Genital evolution: mit 18786), Florida Fish & Wildlife Conservation Commission (partic- why are females still understudied? PLoS Biol. 12:e1001851. ularly N. Gordon), International Fund for Animal Welfare (particularly Arnqvist, G. 1998. Comparative evidence for the evolution of genitalia by M. Niemeyer), New Zealand Common Dolphin Project (particularly K. sexual selection. Nature 393:784–786. Stockin and C. Lea), National Oceanic and Atmospheric Administra- Berdnikovs, S., M. Bernstein, A. Metzler, and R. Z. German. 2007. Pelvic tion’s National Marine Fisheries Service (particularly E. Fougeres and growth: ontogeny of size and shape sexual dimorphism in rat pelves. J. M. Garron), National Oceanic and Atmospheric Administration’s South- Morph. 268:12–22. west Fisheries Science Center (particularly K. Danil and S. Chivers), Berta, A., J. L. Sumich, and K. M. Kovacs. 2015 Marine mammals: evo- North Carolina State University CMAST (particularly V. Thayer and lutionary biology. 3rd ed. Elsevier Academic Press, Amsterdam, The J. Sullivan), Oregon State University (particularly J. Rice), Port Eliza- Netherlands. beth Museum/Bayworld (particularly S. Plon¨ and G. Hofmeyr), Texas Blomberg, S. P., T. Garland Jr., and A. R. Ives. 2003. Testing for phylogenetic Marine Mammal Stranding Network (particularly H. Whitehead and S. signal in comparative data: behavioral traits are more labile. Evolution Piwetz), The Marine Mammal Center (particularly F. Gulland, L. Rust, 57: 717–745. and B. Halaska), University of North Carolina Wilmington (particularly Bookstein, F. L. 1991. Morphometric tools for landmark data. Geometry and W. McLellan), and Virginia Aquarium Stranding Response Program (par- biology. Cambridge Univ. Press, New York, NY. ticularly K. Phillips) for providing specimens. We are grateful to the many Bravo, P. W., J. Moscoso, C. Ordonez, and V. Alarcon. 1996. Transport of interns and volunteers in the Marine Mammal Behavioral Ecology Group spermatozoa and ova in female alpaca. Anim. Reprod. Sci. 43:173– at Texas A&M University at Galveston who assisted with dissections of 179. specimens. We thank S. Theis-Damian and N. Pentyliuk for reviewing lit- Brennan, P. L. R. 2016. Studying genital coevolution to understand intromit- erature and compiling data on cetacean body sizes and masses. We thank tent organ morphology. Integr. Comp. Biol. 56:669–668.

EVOLUTION 2017 11 DARA N. ORBACH ET AL.

Brennan, P. L. R., and R. O. Prum. 2015. Mechanisms and evidence of gen- Lockwood, C. A., J. M. Lynch, and W. H. Kimbel. 2002. Quantifying tem- ital coevolution: the roles of natural selection, mate choice and sexual poral bone morphology of great apes and humans: an approach using conflict. Cold Spring Harb. Perspect. Biol. 7:a017749. geometric morphometrics. J. Anat. 201: 447–464. Brennan, P. L. R., R. O. Prum, K. G. McCracken, M. D. Sorenson, R. E. Macagno, A. L. M., A. Pizzo, H. F. Parzer, C. Palestrini, A. Rolando, and Wilson, and T. R. Birkhead. 2007. Coevolution of male and female A. P. Moczek. 2011. Shape—but not size—codivergence between male genital morphology in waterfowl. PLoS ONE 2:e418. and female copulatory structures in Onthophagus beetles. PLoS ONE Brownell, R. L. Jr., and K. Ralls. 1986. Potential for sperm competition in 6:e28893. baleen whales. Pp. 97–112 in G. P. Donovan, ed. Behaviour of whales Madec, L., and A. Guiller. 1994. Geographic variation of distal genitalia in the in relation to management. Special issue 8 report. International Whaling landsnail Helix aspersa (Mollusca: Gastropoda). J. Zool. 233:215–231. Commission, Cambridge, U.K. Masly, J. P. 2012. 170 years of “lock-and-key”: genital mor- Clarke, R., O. Paliza, and A. L. Aguayo. 1994. Sperm whales of the southeast phology and reproductive isolation. Int. J. Evol. Biol. 2012. Pacific. Pp. 93–224 in G. Pilleri, ed. Investigations on cetacean, part VI, https://doi.org/10.1155/2012/247352. Vol. 25. Museum of Natural History, Paciano, Italy. McGowen, M. R., M. Spaulding, and J. Gatesy. 2009. Divergence date esti- Dines, J. P., E. Otarola-Castillo,´ P. Ralph, J. Alas, T. Daley, A. D. Smith, mation and a comprehensive molecular tree of extant cetaceans. Mol. and M. D. Dean. 2014. Sexual selection targets cetacean pelvic bones. Phylogenet. Evol. 53:891–906. Evolution 68:3296–3306. Mitteroecker, P., P. Gunz, M. Bernhard, K. Schaefer, and F. L. Bookstein. Drake, A. G., and C. P. Klingenberg. 2008. The pace of morphological change: 2004. Comparison of cranial ontogenetic trajectories among great apes historical transformation of skull shape in St. Bernard dogs. Proc. R. Soc. and humans. J. Hum. Evol. 46:679–698. Lond. B. 275:71–76. Norris, K. 1961. Standardized methods for measuring and recording data on El-Banna, A. A., and E. S. E. Hafez. 1972. The uterine cervix in mammals. the smaller cetaceans. J. Mamm. 42:471–476. Am. J. Obst. Gyn. 112:145–164. Orbach, D. N., C. D. Marshall, B. Wursig,¨ and S. L. Mesnick. 2016. Varia- Evans, A., and G. D. Sanson. 2003. The tooth of perfection: functional and spa- tion in female reproductive tract morphology of the common bottlenose tial constraints on mammalian tooth shape. Biol. J. Linn. Soc. 78:173– dolphin (Tursiops truncatus). Anat. Rec. 299:520–537. 191. Orbach, D. N., C. D. Marshall, S. L. Mesnick, and B. Wursig.¨ 2017a. Patterns Evans, J. P., C. Gasparini, G. I. Holwell, I. W. Ramnarine, T. E. Pitcher, and of cetacean vaginal folds yield insights into functionality. PLoS ONE A. Pilastro. 2011. Intraspecific evidence from guppies for correlated 12:e0175037. patterns of male and female genital trait diversification. Proc. R. Soc. Orbach, D. N., D. Kelly, M. Solano, and P. L. R. Brennan. 2017b. Genital in- Lond. B. 278:2611–2620. teractions during simulated copulation amongst marine mammals. Proc. Foth, C., B. P. Hedrick, and M. D. Ezcurra. 2016. Cranial ontogenetic variation R. Soc. B. 284. https://doi.org/10.1098/rspb.2017.1265. in early saurischians and the role of heterochrony in the diversification Orme, D. 2013. The caper package: comparative analysis of phylogenetics of predatory dinosaurs. PeerJ 4:e1589. and evolution in R. R package version 5. Fox, C. H., B. Johnson, J. Whiting, and P. P. Roller. 1985. Fromaldehyde Pabst, D. A., S. A. Rommel, and W. A. McLellan. 1998. Evolution of ther- fixation. J. Histochem. Cytochem. 35:845–853. moregulatory function in cetacean reproductive systems. Pp. 379–397 in Garnier, S., F. Magniez-Jannin, J. Y. Rasplus, and P. Alibert. 2005. When J. G. M. Thewissen, ed. The emergence of whales: evolutionary patterns morphometry meets genetics: inferring the phylogeography of Carabus in the origin of cetacean. Springer, New York, NY. solieri using Fourier analyses of pronotum and male genitalia. J. Evol. Pendergrass, P.B., C. A. Reeves, M. W. Belovicz, D. J. Molter, and J. H. White. Biol. 18:269–280. 2000. Comparison of vaginal shapes in Afro-American, Caucasian and Genevcius, B. C., D. S. Caetano, and C. F. Schwertner. 2017. Rapid differ- Hispanic women as seen with vinyl polysiloxane casting. Gyn. Obstet. entiation and asynchronous coevolution of male and female genitalia in Invest. 50:54–59. stink bugs. J. Evol. Biol. 30:461–473. Perez, S. I., V. Bernal, and P. N. Gonzalez. 2006. Differences between sliding Hedrick, B. P. and P. Dodson. 2013. Lujiatun psittacosaurids: understanding semi-landmark methods in geometric morphometrics, with an applica- individual and taphonomic variation using 3D geometric morphometrics. tion to human craniofacial and dental variation. J. Anat. 208:769–784. PLoS ONE 8:e69265. Perrin, W. F., and G. P. Donovan. 1984. Report of the workshop. Pp. 97–134 in Heinen-Kay, J. L., and R. B. Langerhans. 2013. Predation-associated diver- W. F. Perrin, R. L. Brownell, and D. P. De Master, eds. Reproduction in gence of male genital morphology in a livebearing fish. J. Evol. Biol. whales, dolphin and porpoises. Special issue report 6. Rep. International 26:2135–2146. Whaling Commission, Cambridge, UK. Hosken, D. J., and P. Stockley. 2004. Sexual selection and genital evolution. Perrin, W. F., and S. L. Mesnick. 2003. Sexual ecology of the spinner dol- Trends Ecol. Evol. 19:87–93. phin, Stenella longirostris: geographic variation in mating system. Mar. House, C. M., Z. Lewis, D. J. Hodgson, N. Wedell, M. D. Sharma, J. Hunt, and Mamm. Sci. 19:462–483. D. J. Hosken. 2013. Sexual and natural selection both influence male Pizzo, A., D. Mercurio, C. Palestrini, A. Roggero, and A. Rolando. 2006. Male genital evolution. PLoS ONE 8:e63807. differentiation patterns in two polyphenic sister species of the genus Kamilar, J. M., and N. Cooper. 2013. Phylogenetic signal in primate Onthophagus Latreille, 1802 (Coleoptera: Scarabaeidae): a geometric behaviour, ecology and life history. Phil. Trans. R. Soc. B. 368. morphometric approach. J. Zoo. Syst. Evol. Res. 44:54–62. https://doi.org/10.1098/rstb.2012.0341. Polihronakis, M. 2006. Morphometric analysis of intraspecific shape varia- Kenagy, G. J., and S. C. Trombulak. 1986. Size and function of mammalian tion in male and female genitalia of Phyllophaga hirticula (Coleoptera: testes in relation to body size. J. Mamm. 67:1–22. Scarabaeidae: Melolonthinae). Ann. Entomol. Soc. Am. 99:144–150. Kress, A., and L. Mardi. 1992. Postnatal development of the cervical epithe- Puniamoorthy, N., M. Kotrba, and R. Meier. 2010. Unlocking the “Black lium in the Mongolian gerbil. Anat. Rec. 233:121–134. box”: internal female genitalia in Sepsidae Diptera evolve fast and are Langerhans, R. B., C. A. Layman, and T. J. DeWitt. 2005. Male genital size species-specific. BMC Evol. Biol. 10: 275. reflects a tradeoff between attracting mates and avoiding predators in two R Core Development Team. 2017. R: a language and environment for statistical live-bearing fish species. Proc. Nat. Acad. Sci. USA. 102:7618–7623. computing. R Foundation for Statistical Computing, Vienna, Austria.

12 EVOLUTION 2017 EVOLUTION OF FEMALE CETACEAN GENITAL SHAPE

Ramm, S. A. 2007. Sexual selection and genital evolution in mam- Simmons, L. W., C. M. House, J. Hunt, and F. Garc´ıa-Gonzalez.´ 2009. Evo- mals: a phylogenetic analysis of baculum length. Am. Nat. 169:360– lutionary response to sexual selection in male genital morphology. Curr. 369. Biol. 19:1442–1446. Reinhardt, K. 2010. Natural selection and genital variation: a role for the Singleton, M. 2002. Patterns of cranial shape variation in Papionini (Primates: environment, parasites and sperm ageing? Genetica 138:119–127. Cercopithecinae). J. Hum. Evol. 42:547–578. Rohlf, F. J. 2006. tpsDig, version 2.10. Department of Ecology and Evolution, Sullivan, R. M., S. W. Calhoun, and I. F. Greenbaum. 1990. Geographic State University of New York, Stony Brook, NY. variation in genital morphology among insular and mainland populations Sherratt, E., D. J. Gower, C. P. Klingenberg, M. Wilkinson. 2014. Evolution of Peromyscus maniculatus and Peromyscus oreas. J. Mammal. 71:48– of cranial shape in caecilians (Amphibia: Gymnophiona). Evol. Biol. 58. 41:528–545. von Cramon-Taubadel, N., B. C. Frazier, and M. M. Lahr. 2007. The prob- Showalter, I., B. Todd, and P. L. R. Brennan. 2013. Intraspecific and interspe- lem of assessing landmark error in geometric morphometrics: theory, cific variation of the vagina in two species of water snakes. Biol. J. Linn. methods, and modifications. Am. J. Phys. Anthropol. 134: 24–35. Soc. Lond. 111:183–191. Yassin, A., and V. Orgogozo. 2013. Coevolution between male and female Simmons, L. W. 2014. Sexual selection and genital evolution. Austral. Ento- genitalia in the Drosophila melanogaster species subgroup. PLoS ONE mology 53:1–17. 8:e57158. Simmons, L. W., and R. C. Firman. 2014. Experimental evidence for the Zelditch, M. L., D. L. Swiderski, and H. D. Sheets. 2012. Geometric morpho- evolution of the mammalian baculum by sexual selection. Evolution metrics for biologists: a primer. 2nd ed. Academic Press, Amsterdam, 68:276–283. The Netherlands. Simmons, L. W., and F. Garcia-Gonzalez. 2011. Experimental coevo- Associate Editor: A. Evans lution of male and female genital morphology. Nat. Commun. 2: 374. Handling Editor: M. Noor

Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s website:

Table S1. Biological data on the body lengths or masses of specimens, males, testes, neonates, and mothers. Table S2. Error analysis for individual landmarks using the method developed by Singleton (2002). Table S3. Error analysis for whole specimens using the method developed by Lockwood et al. (2002). Table S4. Principal components 1–4 for all specimens included in the study. Table S5. Principal components for only harbor porpoise (Phocoena phocoena) data. Table S6. Principal components for only common bottlenose dolphins (Tursiops truncatus) data. Table S7. Principal components for sexually mature specimens only. Table S8. Principal components analysis of mean phylogenetic signal. Figure S1. Images of 61 specimens used for GM analyses. Figure S2. Scheme of semilandmark and curve positioning and orientation for trace in TPS.

EVOLUTION 2017 13